Creep behaviour of AZ61 magnesium alloy at low

stresses and intermediate temperatures

V. Srivastava, K. R. McNee, G. W. Greenwood and H. Jones
Tensile creep response was investigated for AZ61 alloy (Mg 6
.
4Al 0
.
9Zn 0
.
2Mn, wt-%) of mean linear
intercept grain size ~25 mm at stresses in the range 0
.
9 4 MPa over the temperature range 250 346 C. Bingham
behaviour is obtained with strain rate e
.
under stress r given by e
.
8r2r
o
with a threshold stress r
o
decreasing from
1
.
25 MPa at 210 C to ~0
.
5 MPa at 346C, which is similar to earlier work on pure magnesium. The corresponding
Arrhenius plot of log (T de
.
/dr) versus T
1
indicates an activation energy comparable with that expected for the grain
boundary self-diffusion coefcient D
B
, and values of D
B
d (where d is the effective grain boundary thickness) derived
from the Coble equation are also similar to those for pure magnesium. Grain elongation in the direction of the tensile
stress is also consistent with the key indicative feature of diffusional creep: deposition of material at grain
boundaries nearly transverse to the axis of tensile stressing. Strain rates versus stress are shown to be continuous
with published results for superplastic ow of AZ61 at comparable temperatures but higher stresses. MST/5838
Keywords: Magnesium alloys, Creep
The authors are in the Department of Engineering Materials, University of Shefeld, Shefeld S1 3JD, UK
(k.a.burton@shefeld.ac.uk). Manuscript received 14 April 2003; accepted 14 July 2003.
# 2004 IoM Communications Ltd. Published by Maney for the Institute of Materials, Minerals and Mining.
Introduction
The creep behaviour of many magnesium alloys at low
stresses and intermediate temperatures remains unexplored,
though early evidence exists for the operation of diffusional
creep mechanisms in several cases,
1 1 1
which is reinforced by
more recent supporting observations.
1 2 1 5
The present work
examines the behaviour of a wrought magnesiumalloy AZ61
with a grain size ~25 mm, which has been reported to exhibit
superplasticbehaviour between 0
.
57 and 0
.
73T
m
and at strain
rates between 10
26
and 10
23
s
21
, with a change in activation
energy from 90 (or 85 kJ) mol
21
below 0
.
64T
m
(or 0
.
7T
m
) to
143 (or 124) kJ mol
21
above it.
1 6 2 0
Experimental
Material of composition Mg 6
.
4Al 0
.
9Zn 0
.
2Mn (wt-%)
was supplied by Magnesium Elektron Limited in the form
of hot rolled plate 10 mm thick. Flat specimens for creep
were machined with gauge length 25 mm parallel to the
rolling direction and gauge width 10 mm, with their at
faces polished to a 1 mm nish before applying random
scratch marks. Tensile creep tests were carried out at
250 346C with applied stresses between 0
.
9 and 4 MPa,
using a Denison machine. The temperature was held within
1K during testing, and the temperature difference along
the gauge length was less than 2 K. Other details of the
experimental procedure were as given by Villa Fanjul et al.
2 1
The creep strain, monitored by a linear transducer, was
conrmed by measuring the gauge length with a travelling
microscope before and after each test. The microstructure
after creep comprised equiaxed a-Mg grains with a small
volume fraction of presumed Mg
1 7
(Al,Zn)
1 2
particles a few
micrometres in size at grain boundaries. The mean linear
intercept matrix grain size was determined on as received
material, after heat treatment for 5 h at 325C, and in the
gauge length and undeformed grip areas of the creep tested
samples. The measurements were made normal and parallel
to the direction of tensile creep load applicationto check for
any developing grain elongation during creep.
Results
The test conditions and resulting incremental strains with
associated secondary creep rates are set out in Table 1. A
typical strain versus time curve (for test 2 at 303C) is shown
in Fig. 1. Signicant primary creep was exhibited only by
Table 1 Test conditions and resulting incremental strains
e and associated secondary creep rates e
.
Test No.
Stress
s, MPa Duration, h
Temperature,
C e, %
e
.
,
10
29
s
21
1
.
1 4
.
00 5 250 0
.
20 138
1
.
2 2
.
67 16 0
.
25 41
.
6
1
.
3 1
.
78 30 0
.
13 12
.
0
1
.
4 0
.
89 67 0
.
03 1
.
7
1
.
5 0
.
89 16 300 0
.
40 72
.
7
1
.
6 2
.
20 5 330 2
.
10 1320
1
.
7 1
.
33 15 3
.
40 528
Total 6
.
51
2
.
1 1
.
80 21 303 2
.
30 368
2
.
2 2
.
70 30 7
.
10 624
2
.
3 0
.
90 113 1
.
70 41
.
8
2
.
4 1
.
10 72 1
.
90 71
.
9
Total 13
.
0
3
.
1 0
.
90 81 300 1
.
40 46
.
6
3
.
2 1
.
35 22 303 1
.
10 146
3
.
3 1
.
35 7 317 1
.
00 423
3
.
4 0
.
90 16 1
.
12 200
3
.
5 0
.
90 21 333 3
.
30 443
3
.
6 1
.
35 3 1
.
00 886
3
.
7 1
.
35 11 346 5
.
20 1490
3
.
8 0
.
90 2 0
.
40 931
3
.
9 1
.
80 3 2
.
60 2520
3
.
10 2
.
25 0
.
3 0
.
40 3550
3
.
11 0
.
90 24 320 2
.
20 263
3
.
12 1
.
35 4 1
.
20 529
3
.
13 1
.
80 16 4
.
30 729
3
.
14 2
.
25 5 317 1
.
80 1030
3
.
15 1
.
80 1 319 0
.
30 684
3
.
16 1
.
80 15 289 0
.
60 166
3
.
17 1
.
35 215 4
.
90 73
.
5
3
.
18 1
.
80 17 0
.
90 147
3
.
19 2
.
25 21 1
.
50 214
3
.
20 1
.
35 69 1
.
60 69
.
4
3
.
21 0
.
90 26 0
.
14 18
.
1
Total 31
.
2
42 Materials Science and Technology January 2004 Vol. 20 DOI 10.1179/026708304225010334
the initial condition for this test (28 h at 0
.
9 MPa), the
subsequent stages at 1
.
8, 2
.
7 and 0
.
9 MPa showing minimal
primary components. Figure 2 shows that secondary creep
rates exhibit a linear dependence on applied stress between
1 and 4 MPa at all temperatures, with a temperature
dependent threshold stress s
o
. Values of strain rate per unit
stress de
.
/ds and of s
o
from Fig. 2 are given in Table 2 along
with other parameters. Figures 3 and 4 show Arrhenius
plots of T de
.
/ds and s
o
versus reciprocal temperature.
Table 3 shows the mean linear intercept grain size L

parallel
and normal to the rolling direction and direction of creep
loading. The initial equiaxed grain structure did not change
signicantly from its initial L

of 25 mm in a 5 h heat treat-
ment at 325C or during tests 1 and 2 (a small increase to
about 30 mm resulted from test 3). More signicantly, after
creep, the grain dimension in the direction of creep loading
in the gauge length was 10 15% larger than in the trans-
verse direction, whereas grains remained equiaxed in the
grip areas. For the 13% strain of sample 2 and the 8% strain
of sample 3, the measured average grain aspect ratio is
sufcient to account for the measured sample strain. For the
31% strain of sample 3, however, the measured aspect ratio
is not sufcient to account for the measured strain, indi-
cating that some additional process has made an appreci-
able contribution to this larger strain. Finally, there was
evidence from the reorientation of scratch marks made
1 Strain versus time curves for test 2 at 303 C for stresses
indicated
j 250C; m 289C; d 302C;
h
318C; 6 330C; h 333C;
s 346C
2 Steady state strain rate e
.
versus stress r
Table 2 Threshold stress r
o
and steady state strain rate per unit stress de
.
/dr versus test temperature T for AZ61 from
the data in Fig. 2; values of D
B
d derived via equation (4) from the de
.
/dr results are also shown together with
the Coble creep derived values of D
B
d obtained by Crossland
23
for pure Mg
Test temperature, C Threshold stress s
o
, MPa de
.
/ds, 10
213
Pa
21
s
21
Equivalent D
B
d, m
3
s
21
D
B
d for Mg, m
3
s
21
(Ref. 23)
D
B
d(AZ61)
D
B
d(Mg)
250 1
.
250
.
60 0
.
450
.
11 2
.
44610
20
2
.
61610
220
0
.
94
289 0
.
820
.
11 1
.
530
.
10 8
.
74610
20
1
.
39610
220
0
.
63
303 0
.
840
.
16 3
.
090
.
26 1
.
81610
19
2
.
41610
219
0
.
75
318 0
.
510
.
11 5
.
710
.
40 3
.
43610
19
4
.
20610
219
0
.
82
330 0
.
270
.
40 6
.
741
.
76 4
.
13610
19
6
.
42610
219
0
.
64
346 0
.
500
.
15 19
.
71
.
8 1
.
24610
18
1
.
10610
218
1
.
12
3 Logarithm of temperature compensated steady state
strain rate per unit stress Tde/dr versus inverse test
temperature T from data in Table 2
4 Logarithm of threshold stress r
o
versus inverse test
temperature T from data in Table 2
Srivastava et al. Creep behaviour of AZ61 magnesium alloy 43
Materials Science and Technology January 2004 Vol. 20
parallel to the direction of loading that some rotation of
grains did indeed occur during creep. An example is shown
in Fig. 5 for sample 2.
Discussion
The results of Fig. 2 are characteristic of Bingham
creep behaviour, where e
.
8(s2s
o
), and s
o
ranges from
1
.
250
.
6 MPa at 250C to 0
.
500
.
15 MPa at 346C.
Similar behaviour has been reported for several
metals
8,2 2 2 6
and some ceramics
2 7 30
under comparable
conditions. The results for pure magnesium,
2 3
for example,
range from 1
.
5 MPa at 152C to 0
.
16 MPa at 323C and
show a reasonable t with the empirical equation
s
o
~
A

LL
exp
U
RT

: : : : : : : : : : : : (1)
where A and U are material constants. The results in Fig. 4
give A~0
.
150
.
02 N m
21
and U~23
.
15
.
7 kJ mol
21
for
AZ61 with L

~25 mm, comparedwith A~0

.
040
.
01 N m
21
and U~23 kJ mol
21
for Mg (Ref. 23) with L

~15 mm.
The activation enthalpy of 1069 kJ mol
21
yielded by
the temperature compensated strain rate per unit stress
results in Fig. 2 is ~0
.
7 of the activationenthalpy for lattice
diffusion of Mg in Mg.
32 35
which would be consistent with
a major role for grain boundarydiffusion in determining the
creep rate. This is also consistent with the ndings of
Crossland and Jones
2 3
for pure magnesium, that grain
boundary diffusion creep (Coble creep) dominated over the
ranges 152<T<323C and 15<L

<104 mm at low stresses

comparable with the conditions in the present experiments
for AZ61. Crosslands activation energy for this low stress
creep of pure magnesium (105 kJ mol
21
) is also similar to
our value for AZ61. The relevant equations are then
(d_ ee=ds)
L
~
12VD
L
d
2
kT
: : : : : : : : : : : (2)
and
(d_ ee=ds)
B
~
48VD
B
d
d
3
kT
: : : : : : : : : : : (3)
for latticediffusion(Nabarro Herring)
36,37
andgrainbound-
ary diffusion (Coble) creep,
38
respectively. Taking V~
2
.
33610
22 9
m
3
atom
21
, k~1
.
38610
22 3
J atom
21
K
21
,
d~1
.
76L

and D
L
~5
.
9610
21 6
m
2
s
21
(Refs. 32, 35), equa-
tion (2) predicts (de
.
/ds)
L
~1
.
0610
21 4
m
2
N
21
s
21
at our
highest test temperature, 346C, which is less than 1% of the
measured value, implying a negligible contribution from
Nabarro Herring creep. The present results for de
.
/ds can
be compared with those of Crossland by substituting them
for (de
.
/ds)
B
in equation (3) to yield
D
B
d~(d
3
kT=48V)d_ ee=ds : : : : : : : : : (4)
at each test temperature. The resulting D
B
d values are
included in Table 2 and plotted as log (D
B
d) versus 1/T
in Fig. 6, yielding D
O B
d~(7
.
71
.
6)610
21 0
m
3
s
21
and
Q
B
~1069 kJ mol
21
for AZ61. This result is in excellent
agreement with the corresponding values obtained by
Crossland for pure magnesium, D
O B
d~8610
21 0
m
3
s
21
and Q~105 kJ mol
21
. The resulting values of D
B
d for our
test temperatures are given for comparison in Table 2.
Therefore, it appears that the content of Al and Zn in AZ61
5 Evidence that grain rotation occurred during test 2
Table 3 Mean linear intercept grain size L

for different conditions

Orientation relative to rolling direction and axis of creep loading Parallel L

P
, mm Normal L

n
, mm L

P
/L

n
As received 21
.
41
.
3 21
.
91
.
1 0
.
98
After 5 h at 325C 25
.
51
.
7 25
.
01
.
3 1
.
02
Sample 2 at end of test
in grip area 28
.
92
.
4 29
.
32
.
3 0
.
99
in gauge length (at 13% strain) 22
.
61
.
8 20
.
62
.
2 1
.
10
Sample 3 at end of test
in grip area 25
.
91
.
5 26
.
21
.
6 0
.
99
in gauge length (at 8% strain) 26
.
21
.
3 23
.
41
.
8 1
.
12
in gauge length (at 31% strain) 32
.
72
.
5 28
.
62
.
2 1
.
15
6 Dependence on reciprocal test temperature of loga-
rithm of D
B
d given by equation (4) from measured de/
dr for AZ61 from Table 2
44 Srivastava et al. Creep behaviour of AZ61 magnesium alloy
Materials Science and Technology January 2004 Vol. 20
has a negligible effect on the magnitude of grain boundary
diffusivity indicated by the Coble creep process.
The implication that Coble creep with a threshold stress
controls the creep process operating under the conditions
studied is conrmed by the increase in aspect ratio of grains
along the direction of applied stress from 1
.
000
.
02 as
received and in the grip areas to 1
.
10 1
.
15 as a result of
creep (Table 3). Some rotation of grains evident in Fig. 5 is
a further expected consequence of the diffusional creep
deformation of an assembly of quasi-equiaxedgrains which
remain in mutual contact during transfer of matter to and
from grain faces.
Finally, it is of interest to compare the ndings of
the present creep work with the results for superplastic
deformation of AZ61 with a similar grain size at similar
stresses and temperatures. Watanabe et al.
1 6
used AZ61
with a grain size (L

:20 mm) similar to that of our material

in the comparable temperature range (250 400C). Their
stress range extended up to ~70 MPa, however, and
stresses below our maximum of 4 MPa were explored only
at their highest temperatures (375 and 400C), with the
operativeactivationenergy90 kJ mol
21
at (s2s
o
)~25 MPa,
similar to our 1069 kJ mol
21
, in the range 250 300C.
Grain size was also found to be relatively stable up to about
380C, consistent with our ndings. The subsequent publi-
cation by Kim et al.
2 0
used AZ61 with L

~5 and 9 mm
at 325 420C, again with stresses down to our maximum
of 4 MPa reached only at the highest temperature. At the
lower temperatures and s~22 MPa, activation energy was
near to 90 kJ mol
21
, similar to our 1069 kJ mol
21
, and
grain size was relatively stable up to nearly 400C.
Figure 7a e shows strain rate versus stress for the ve
lowest temperatures employed by Watanabe et al. and Kim
et al. in their superplasticity work, for comparison with our
creep data at lower stresses for the same or near equivalent
temperatures. The continuity between our creep and their
a 250C; b 275/289C; c 300/303C; d 325/330C; e 350/346C; j Watanabe et al.
16
(e~0
.
1); m Kim et al.
20
(L

~5 mm); . Kim et al.

20
(L

~6
.
7 mm); d present work
7 Strain rate versus stress for ve lowest temperatures used by Watanabe et al.
16
and Kim et al.
20
for their superplasti-
city work on AZ61, for comparison with creep data at lower stresses and similar temperatures
Srivastava et al. Creep behaviour of AZ61 magnesium alloy 45
Materials Science and Technology January 2004 Vol. 20
superplastic strain rates is notable. This continuity is
reinforced by the corresponding plot of temperature
compensated strain rate versus s/E, where E is Youngs
modulus, in Fig. 8, using our value of activation energy,
106 kJ mol
21
. The increase in stress exponent at s/E>10
23
is indicative, however, that the mechanism of superplastic
ow involves more than diffusional creep.
Conclusions
1. Steady state creep of AZ61 alloy at 250<T<346C
and 0
.
9<s<4 MPa and an intercept grain size L

of
~25 mm exhibits Bingham behaviour, with s
o
decreasing
with increasing T from ~1
.
25 MPa at 250C to ~0
.
5 MPa
at 346C, in good accord with earlier data for pure mag-
nesium under similar conditions.
2. The Arrhenius plot of log (Tde
.
/ds) versus T
21
yields
an activation energy similar to that expected for control of
creep by grain boundary diffusion, and consistent with the
expectation that Coble rather than Nabarro Herring dif-
fusion creep should dominate for the conditions of the test.
The resulting values of D
B
d obtained from measurements
of de
.
/ds values closely match those obtained earlier by
Crossland for pure magnesium, and give D
O B
d~7
.
76
10
21 0
m
3
s
21
with Q
B
~1069 kJ mol
21
for the range 250<
T<346C.
3. Measurements of grain aspect ratio showelongationin
the direction of application of the tensile and contraction in
the transverse direction dependent on strain. There was no
such elongation in the unstressed grip areas after creep or in
the material, as heat treated or as received. This is consistent
with the stress directed diffusive deposition of material at
boundaries under normal stress, a key indicator of the opera-
tion of a diffusional creep mechanism.
4. The results for strain rate versus stress are shown to be
continuous with those of Watanabe et al. and Kim et al. for
a superplastic ow of AZ61 at comparabletemperaturesbut
higher stresses.
Acknowledgements
This work forms part of a PhD research programme by VS
at the University of Shefeld funded by a project student-
ship funded by a UK EPSRC research grant award (GR/
N00296), combined with Universityof Shefeld fee bursaries.
References
1. r. l. squires, r. t. weiner and m. phillips: J. Nucl. Mater.,
1963, 8, 77 80.
2. j. e. harris and r. b. jones: J. Nucl. Mater., 1963, 10, 360 362.
3. j. h. day: J. Nucl. Mater., 1964, 11, 249 251.
4. r. b. jones: Nature, 1965, 207, 70.
5. v. j. haddrell: J. Nucl. Mater., 1966, 18, 231 232.
6. r. b. jones: in Quantitative relations between properties and
microstructure, (ed. D. G. Brandon and A Rosen), 343 349;
1969, Jerusalem, Israel Universities Press.
7. i. g. crosslandandj. c. wood: Philos. Mag., 1975, 31, 1415 1419.
8. i. g. crossland: in Physical metallurgy of reactor fuel
elements, (ed. J. E. Harris and E. C. Sykes), 66 71; 1975,
London, The Metals Society.
9. g. f. hines, v. j. haddrell and r. b. jones: in Physical
metallurgy of reactor fuel elements, (ed. J. E. Harris and E. C.
Sykes), 72 78; 1975, London, The Metals Society.
10. i. g. crossland and r. b. jones: Met. Sci., 1977, 11, 504 508.
11. s. s. vagarali and t. g. langdon: Acta Mater., 1981, 29,
1969 1982.
12. k. r. mcnee, h. jones and g. w. greenwood: in Creep and
fracture of engineering materials and structures, (ed J. D.
Parker), 3 13; 2001, London, The Institute of Materials.
13. k. r. mcnee, g. w. greenwood and h. jones: Philos. Mag. A,
2002, 82, 2773 2790.
14. k. r. mcnee, g. w. greenwood and h. jones: Scr. Mater., 2002,
46, 437 439.
15. k. r. mcnee, g. w. greenwood and h. jones: Scr. Mater., 2002,
47, 619 623.
16. h. watanabe, t. mukai, m. kohzu, s. tanabe and k. higashi:
Acta Mater., 1999, 47, 3753 3758.
17. h. tsutsui, h. watanabe, t. mukai, m. kohzu, s. tanabe and
k. higashi: Mater. Trans., JIM, 1999, 40, 931 934.
18. w. j. kim and s. w. chung: Metals and Mater., 2000, 6, 255 259.
19. w. j. kim, s. w. chung, c. w. an and k. higashi: J. Mater. Sci.
Lett., 2001, 20, 1635 1637.
20. w. j. kim, s. w. chung, c. s. chung and d. kum: Acta Mater.,
2001, 49, 3337 3745.
21. f. villa fanjul, s. srimanosaowapak, k. r. mcnee, g. w.
greenwood and h. jones, Z Metallk., 2003, 94, 25 29.
22. d. j. towle and h. jones: Acta Met, 1976, 24, 399 407.
23. i. g. crossland and r. b. jones: Met. Sci., 1977, 11, 504 508.
24. t. sritharan and h. jones: Acta Metall., 1979, 27, 1293 1300.
25. t. sritharan and h. jones: Acta Metall., 1989, 28, 1633 1639.
26. t. sritharan and h. jones: Met. Sci., 1981, 15, 365 368.
27. a. g. crouch: J. Am. Ceram. Soc., 1972, 55, 558 563.
28. k. a. hay, a. g. crouch and r. t. pascoe: in Physical
metallurgy of reactor fuel elements, (ed. J. E. Harris and E. C.
Sykes), 95 98; 1975, London, The Metals Society.
29. b. burton and g. l. reynolds: Acta Metall., 1973, 21, 1073 78.
30. b. burton and g. l. reynolds: in Physical metallurgy of
reactor fuel elements, (ed. J. E. Harris and E. C. Sykes),
87 94; 1975, London, The Metals Society.
31. b. burton: Mater. Sci. Eng., 1972, 10, 9 14.
32. p. g. shewmon and f. n. rhines: Trans. AIME, 1954, 200,
1021 1025.
33. p. g. shewmon: Trans. AIME, 1956, 206, 918 922.
34. j. combronde and g. brebec: Acta Metall., 1971, 19, 1393
1399.
35. s. i. fujikawa: J. Jpn. Inst. Light. Met, 1992, 42, 822 825.
36. f. r. n. nabarro: Report of a conference on The strength of
solids, 75 99; 1948, London, The Physical Society.
37. c. herring: J. Appl. Phys., 1950, 27, 437 445.
38. r. l. coble: J. Appl. Phys., 1963, 34, 1679 1682.
8 Temperature compensated strain rate versus r/E for
superplasticity data
16,20
compared with present creep
data for AZ61 using activation energy for creep
106 9 kJ mol
1
from present work
46 Srivastava et al. Creep behaviour of AZ61 magnesium alloy
Materials Science and Technology January 2004 Vol. 20